 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
Less
obvious work:
|
|
|
|
– |
Changing the learning methodology. Feature
selection,
|
|
|
longer-range
CRF constraints
|
|
|
|
– |
Cleaning up the annotation guidelines with
feedback from
|
|
automatic
extraction results
|
|
|
Longer
term:
|
|
|
|
– |
Integration with larger project (ingestion
into MySQL)
|
|
|
|
– |
Data cleaning (variations on institution
names,
|
|
|
misspellings,
misspacings)
|
|